1. Field of the Invention
The present invention relates to a manufacturing method of a semiconductor device having a circuit in which a thin film transistor (hereinafter referred to as TFT) is formed on a substrate having an insulating surface. More particularly, the present invention is suitable for use in an electro-optical device typically known as a liquid crystal display device having a pixel TFT formed in a display region and a drive circuit formed in the periphery of the display region on the same substrate, and in an electronic equipment mounted with an electro-optical device with this type. Note that throughout the specification of the present invention, the semiconductor device indicates any device utilizing the semiconductor characteristics for functioning, and includes the foregoing electro-optical device and an electronic equipment mounted with the electro-optical device in its category.
2. Description of the Related Art
From the fact that a TFT having an active layer formed of a crystalline silicon film on a substrate having an insulating surface (hereinafter referred as a crystalline silicon TFT) has a high electric field effect mobility, it is possible to form a variety of functional circuits. The above electro-optical device having such functional circuits integrally formed on the same substrate has been developed. As a typical example, the active matrix liquid crystal display device is well known.
In the active matrix liquid crystal device employing the crystalline silicon TFT, a pixel TFT is formed in every pixel of an image display region and a drive circuit is formed in the periphery of the image display region. The drive circuit is composed of a shift resister circuit, a level shifter circuit, a buffer circuit, a sampling circuit, and the like in which a CMOS circuit as a basic circuit is formed. These circuits are formed on the same substrate, and formed into a unity to complete a display device.
The operating conditions of the pixel TFT and the drive circuit are not necessarily the same. From this, the characteristics that are demanded for a TFT is somewhat different. For example, the pixel TFT is demanded to function as a switch device for applying a voltage to a liquid crystal. The liquid crystal is driven by an a.c., thus a method called frame inverse drive is widely adopted. In this method, for the purpose of maintaining an electric charge of a storage capacitor, the characteristic that is demanded for the pixel TFT was to sufficiently lower an off-current value (a drain current that flows during an off-operation of TFT). On the other hand, since a high drive voltage is applied to the buffer circuit of the drive circuit, it was necessary to raise the voltage-resistance of the TFT so that it will not break when a high voltage is applied. Also, in order to make a current drive ability higher, it was necessary to sufficiently secure an on-current value (a drain current that flows during an on-operation of TFT.
However, the point at issue is that the off-current value of the crystalline silicon TFT easily rises. Moreover, similar to an MOS transistor used in an IC and the like, a deterioration phenomenon such as a drop of the on-current value is observed on the crystalline silicon TFT. This mainly results from a hot carrier implantation. It has been considered that the hot carrier generated by a high electric field in the vicinity of the drain triggers the deterioration phenomenon.
As a structure of the TFT to reduce the off-current value, an LDD region (Lightly Doped Drain) is known. In this structure, there is provided a region that is added with an impurity element at a low concentration between a channel forming region and a source region or a drain region which is formed by adding an impurity element at a high concentration, and this region is called the LDD region.
For example, Japanese Patent No. 2564725 discloses a method of manufacturing a TFT having an LDD region. In this method, a gate insulating film is formed widely in the direction of a channel width from a gate electrode, and an insulating film thinner than the gate insulating film is further formed beside this gate insulating film. Then by utilizing the difference in the thickness between this insulating film and the gate insulating film, an LDD region is formed in a semiconductor film between the end portion of the gate electrode and a source or drain region.
Moreover, as means for preventing deterioration caused by a hot carrier, a so-called GOLD (Gate-drain Overlapped LDD) structure is known in which the LDD region is arranged so as overlapping the gate electrode via the gate insulating film. With a structure of this kind, the high electric field in the vicinity of a drain is relaxed to prevent hot carrier implantation, which is effective for, prevention of the deterioration phenomenon. For example, though a GOLD structure formed by a side wall which is formed of silicon is disclosed by Mutuko Hatano, Hajime Akimoto and Takeshi Sakai, in “IEDM97 TECHNICAL DIGEST, p 523-526, 1997,” it has been confirmed that an extremely excellent reliability can be achieved when compared with TFTs of other structures.
It had been preferred that the introduction of an impurity element to the semiconductor layer, which is used to form an impurity region such as the LDD, region, the source or drain region of the TFT having such a structure, be conducted in a self-aligning manner utilizing the gate electrode and the insulating film for a mask provided on the semiconductor layer. Furthermore, in order to reduce the number of masks, a, method (referred as cross dope method in the present invention) has been employed in which once an impurity element of a one conductive type is introduced into the whole surface by utilizing the gate electrode and the insulating film as a mask, an impurity element of a conductive type opposite the one conductive type is introduced into the impurity region of a TFT of either the P-channel TFT or the N-channel TFT at a high concentration.
However, the characteristics being demanded for the pixel TFT and the TFT of the drive circuit such as a shift resist circuit, or a buffer circuit are not necessarily the same. For example, in the pixel TFT, a large reversal bias (negative voltage in an N-channel TFT) is applied to a gate, whereas the TFT of the drive circuit basically does not operate in the reversal bias state. Also, the operating velocity of the pixel TFT can be 1/100 of less than that of the TFT of the drive circuit.
The GOLD structure is highly effective in preventing the deterioration of the on-current value, but on the other hand, there arises a problem in that the off-current becomes higher compared with the structure of a normal LDD. Therefore, the GOLD structure was not a preferred structure for applying to the pixel TFT. Contrarily, although the structure of a normal LDD is highly effective in suppressing the off-current value, it has a low effect in relaxing the electric field in the vicinity of a drain and in preventing deterioration caused by the hot carrier implantation. In this way, in a semiconductor device such as the active matrix liquid crystal display device that has a multiple of integrated circuits of different operating conditions, it was not always desirable to form all the TFTs with the same structure. Such a point of issue like has been revealed as an enhanced characteristic is required for the crystalline silicon TFT in particular, and as the performance demanded of the active matrix liquid crystal display device becomes increased.
Furthermore, although there are several means to decrease the off-current value of the TFT, it was necessary to form a good junction of the channel forming region and the impurity region (LDD region, source region or drain region). In order to do this, the distribution of an impurity element in the interface of the channel forming region and the impurity region that contacts the channel forming region needs to be accurately controlled. However, when the above-mentioned cross dope method is implemented, the distribution of an impurity element in the interface was difficult to control accurately since an impurity element of a one conductive type and an impurity element of a conductive type opposite the one conductive type were introduced into the impurity region of the TFT on one side.
This type of LDD structure is formed by focusing on the characteristic of the N-channel TFT. Many of the P-channel TFTs, which are formed on the same substrate to form the CMOS circuit or the like, were formed as a single drain structure, for the purpose of lessening the number of masks as possible. In this case, there arises a problem in that since phosphorous (P) for forming an LDD in the N-channel TFT is doped into the source or drain region in the P-channel TFT, a defect is formed in the junction with the channel forming region and the off-current value increases.